Engineers at Hangzhou-based DeepSeek have revealed the innovative training techniques used for their viral AI model DeepSeek-R1. Released in January, this open-source model challenges OpenAI's o1 by utilizing reward-based training to overcome traditional computational barriers. In a paper published in Nature, the team emphasizes that "general reasoning represents a long-standing and formidable challenge in artificial intelligence," essential for executing tasks like mathematical problem solving, signaling a major step towards humanlike AI development.